| Introduction | 12
|
| Chapter 1. Philosophical foundation of investigations of cognitive evolution | 18
|
| 1.1. Toward the theory of evolutionary origin of the human thinking | 18
|
| 1.1.1. Einstein’s lesson | 18
|
| 1.1.2. The supertask | 20
|
| 1.1.3. Hume → Kant → Lorenz | 24
|
| 1.1.4. “The unreasonable effectiveness of mathematics in natural sciences” and the problem of principal applicability of the human thinking to cognition of nature | 28
|
| 1.1.5. Internal model and prediction | 30
|
| 1.2. Conceptual theories | 30
|
| 1.2.1. Petr Anokhin’s functional system: the general scheme of animal control system | 31
|
| 1.2.2. Turchin’s theory of metasystem transitions | 35
|
| 1.3. Concluding remarks | 38
|
| Chapter 2. Backgrounds for modeling of cognitive evolution | 43
|
| 2.1. Backgrounds in computer science | 43
|
| 2.2. Biological experiments on “elementary thinking of animals” | 48
|
| Chapter 3. Method of reinforcement learning | 55
|
| 3.1. Reinforcement learning | 55
|
| 3.2. Adaptive critic designs | 58
|
| Chapter 4. Analysis of evolutionary optimization methods | 65
|
| 4.1. Estimation of the rate and effectiveness of evolutionary algorithms | 65
|
| 4.1.1. Quasispecies model | 65
|
| 4.1.2. Qualitative picture of evolution in the quasispecies model | 70
|
| 4.1.3. Stochastic nature of the evolutionary process. The role of neutral selection | 72
|
| 4.1.4. Estimation of rate and efficiency of evolutionary process | 74
|
| 4.1.5. Results of computer simulation | 76
|
| 4.1.6. Comparison of evolutionary search with other methods | 79
|
| 4.1.7. The case of several symbols of optimized chains | 80
|
| 4.1.8. Model of a narrow channel and majority model | 81
|
| 4.1.9. Spin-glass model of evolution | 85
|
| 4.1.10. Conclusions to the estimations of the rate and efficiency of evolution | 90
|
| 4.2. Model of interaction between learning and evolution | 92
|
| 4.2.1. Background studies | 92
|
| 4.2.2. Description of the model | 95
|
| 4.2.3. Results of computer simulation | 99
|
| 4.2.3.1. Scheme and parameters of simulation | 99
|
| 4.2.3.2. Comparison of regimes of pure evolution and evolution combined with learning | 101
|
| 4.2.3.3. Hiding effect | 105
|
| 4.2.3.4. Influence of the learning load on the modeled processes | 110
|
| 4.2.3.5. Probabilistic and deterministic selection | 113
|
| 4.2.3.6. Modeling of Lamarckian evolution | 114
|
| 4.2.4. Comparison with the approach by Hinton and Nowlan | 115
|
| 4.2.5. Conclusion to the model of interaction between learning and evolution | 120
|
| 4.3. Model of imprinting formation by means of learning and evolution | 121
|
| 4.3.1. Model description | 122
|
| 4.3.2. Simulation results | 127
|
| 4.3.2.1. Scheme and parameters of simulation | 127
|
| 4.3.2.2. Comparison of regimes of pure evolution and evolution combined with learning | 129
|
| 4.3.2.3. Hiding effect | 133
|
| 4.3.2.4. Influence of the learning load on the modeled processes | 136
|
| 4.3.3. Discussion of the model of formation of imprinting | 138
|
| Chapter 5. Models of autonomous adaptive agents | 142
|
| 5.1. Adaptive syser: model of a proto-organism that adapts to variable external environment | 142
|
| 5.1.1. Model of sysers | 143
|
| 5.1.1.1. General scheme of sysers | 143
|
| 5.1.1.2. Mathematical description of sysers | 144
|
| 5.1.1.3. Sysers and self-reproducing automata by John von Neumann | 149
|
| 5.1.2. Model of adaptive syser | 150
|
| 5.1.2.1. Biological prototype of adaptive syser | 151
|
| 5.1.2.2. General scheme of functioning of the adaptive syser | 153
|
| 5.1.2.3. Mathematical description of mini- and adaptive sysers | 155
|
| 5.1.3. Conclusion to the model of adaptive syser | 162
|
| Appendix to Section 5.1. Analysis of the dynamic system | 163
|
| 5.2. “Grasshopper”: model of evolutionary origin of goal-directed adaptive behavior | 166
|
| 5.2.1. Model description | 166
|
| 5.2.1.1. Main assumptions of the model | 166
|
| 5.2.1.2. Overview of environment and agents | 167
|
| 5.2.1.3. Agent physiology | 169
|
| 5.2.1.4. Neural network of an agent | 171
|
| 5.2.1.5. Scheme of evolution | 173
|
| 5.2.2. Computer simulation | 173
|
| 5.2.2.1. Parameters of computer simulation | 173
|
| 5.2.2.2. Simulation results | 176
|
| 5.2.3. Analysis of results | 179
|
| 5.2.4. Conclusion to themodel “Grasshopper” | 182
|
| 5.2.5. Development of the model “Grasshopper”: the emergence of the naturally branched hierarchy of goals | 182
|
| 5.3. Biologically inspired model of adaptive searching behavior | 184
|
| 5.3.1. Searching behavior of living and modeled organisms | 184
|
| 5.3.2. The searching behavior of caddisflies larvae. Results of biological experiment | 187
|
| 5.3.3. Model of searching behavior of caddisflies larvae | 189
|
| 5.3.3.1. Description of the main variant of the model | 189
|
| 5.3.3.2. Results of computer simulation | 191
|
| 5.3.3.3. Additional model | 197
|
| 5.3.3.4. Conclusion to the model of searching behavior of caddisflies larvae | 199
|
| 5.3.4. Biologically inspired method of functions optimization | 200
|
| 5.3.5. Conclusion to the biologically inspired method of searching | 206
|
| 5.4. Modeling of searching agent behavior by means of neural gas | 207
|
| 5.4.1. One-dimensional case | 209
|
| 5.4.2. Two-dimensional case | 214
|
| 5.4.3. Conclusion to modeling of searching agent behavior by means of neural gas | 216
|
| 5.5. Chapter summary | 217
|
| Chapter 6. The sketch program for future investigations of cognitive evolution | 221
|
| Chapter 7. The initial steps of modeling of cognitive evolution | 225
|
| 7.1. Model of autonomous agents with several natural needs | 225
|
| 7.1.1. Model of agents with natural needs and motivations | 225
|
| 7.1.1.1. Modeled world | 226
|
| 7.1.1.2. Agent control system | 226
|
| 7.1.1.3. Hierarchy of motivations | 227
|
| 7.1.1.4. Scheme of learning | 230
|
| 7.1.2. Results of computer simulation | 230
|
| 7.1.3. Conclusion to the model of agents with natural needs | 232
|
| 7.2. Model of formation of generalized notions by autonomous agents | 232
|
| 7.3. Models of fish exploratory behavior in mazes | 236
|
| 7.3.1. Short description of biological experiments. Behavior of fish in mazes | 237
|
| 7.3.1.1. Behavior of zebrafish in the cross-shaped maze | 238
|
| 7.3.1.2. Behavior of fish in the maze with 11 arms | 239
|
| 7.3.2. Models of fish movements, accumulation of knowledge, formation of predictions | 240
|
| 7.3.2.1. Model of knowledge acquisition | 240
|
| 7.3.2.2. Model of predictions of future situations | 244
|
| 7.3.3. Hypothetical model of planning of movement in the maze with 11 arms | 247
|
| 7.3.4. Conclusion to models of fish exploratory behavior in mazes | 254
|
| 7.4. Modeling of mechanism of plan formation by New Caledonian crows | 255
|
| 7.4.1. Biological experiment on NC crows | 256
|
| 7.4.2. Mechanism of plan formation | 257
|
| 7.4.2.1. Description of the model | 257
|
| 7.4.2.2. Results of computer simulation | 262
|
| 7.4.3. Discussion of the model of plan formation by NC crows | 264
|
| 7.5. Chapter summary | 266
|
| Chapter 8. Possible applications related to modeling of cognitive evolution | 268
|
| 8.1. Development of scientificpoint of view. Approaches to harmonious development of mankind | 269
|
| 8.2. Elimination of genes of aggressiveness in the evolving population of conflicting agents and idea of a project for the Nobel peace award | 271
|
| 8.3. Agent-based model of transparent market economy | 277
|
| 8.3.1. Description of the model | 278
|
| 8.3.1.1. General scheme of the model | 278
|
| 8.3.1.2. Description of the iterative process | 280
|
| 8.3.2. Results of computer simulation | 282
|
| 8.3.3. Conclusion to the model of transparent market economy | 287
|
| 8.4. Chapter summary | 288
|
| Chapter 9. Interdisciplinary relations of modeling of cognitive evolution | 290
|
| Conclusion | 294
|
|
|
|
|
|
|
Редько Владимир Георгиевич